BRAZIL UPGRADES SYSTEM FOR FAULT MANAGEMENT
CURRENTLY, AES ELETROPAULO, which serves the city of São Paulo, Brazil, experiences approximately 30,000 events or faults per month on its distribution system. An existing corporate system, Gerência de Ocorrências da Distribuição (GOD), is responsible for managing all events and keeps online records of them for the last three years.
Although this system stores detailed information on every event, identifying the probable cause for each event is difficult for a number of reasons. Firstly, the information collected directly from the field is sometimes incomplete or inadequate. Secondly, the system was not designed to easily integrate information provided by external sources, including information from various state-level agencies on weather and climatic conditions such as lightning, rain, wind and air pollution. Finally, the corporate system lacks the statistical methodology to establish relationships between the events and parameters that affect the electrical network.
As a result, AES Eletropaulo launched a research project to integrate data resources that relates field conditions to faults so better measures can be taken to avoid network disturbances.
INTEGRATING DATA SOURCES INTO A SINGLE DATABASE
Data is available from a number of sources:
Event management system, GOD
This system collects and manages event information, from the initial steps to the generation of consolidated reports at various detail levels. Because it runs on a mainframe platform, there is no direct interaction between field personnel and the system; this communication was formerly performed through specific paper forms.
Electrical network management system, GRADE
This system contains the data on network components and operational information, including medium-voltage (MV) and low-voltage (LV) circuits, pole locations, network section cables, protective and switching equipment, location and state (open/closed), consumer location and demand, and results from load-flow calculations. It runs on a mainframe platform but lacks a graphical user interface. All details are geo-referenced, which means they possess a corresponding pair of universal transverse mercator (UTM) coordinates, so the geographical information is already in the database. The GOD system normally issues various queries to the GRADE system to determine circuit topology, state of protective/switching equipment and demand levels at specific locations.
Wind data
In São Paulo, wind and air pollution data are collected and processed by the state agency Companhia de Tecnologia de Saneamento Ambiental (CETESB), which operates 23 local stations located throughout São Paulo's metropolitan area. The wind data available comprises maximum wind speed (m/s) recorded at 1-hour intervals. For each recorded event, the software identifies the station closest to the event location and retrieves the maximum wind speed in the hour prior to the event and the maximum wind speed 24 hours prior to the event.
Air pollution data
CETESB also collects and processes air pollution data in São Paulo city. The data includes hourly values of inhalable particles, measured in µg/m
Rain data
Centro Tecnológico de Hidráulica is the agency responsible for collecting rainfall data at 5-minute intervals in various ranges (0-5 mm/hour, 5-7 mm/hour, etc.). For each recorded event, the software identifies the corresponding square on the rain grid and computes the accumulated rainfall in the preceding 60 minutes and 24 hours.
Lightning data
This information is available from Sistema Meteorológico do Paraná, a regional system. For each detected lightning stroke, the following information is made available: GMT time of stroke, stroke intensity and parameters of the ellipse that most likely contain the stroke location. For each recorded event, the software determines if the event location belongs to the ellipse of any recorded stroke, both at the time of the event and in the preceding 24 hours. (The software automatically converts GMT time to local time, accounting for energy-saving summer time corrections if applicable.)
MAIN FUNCTIONAL ASPECTS
The main functional aspects of the event/fault research project were:
Field data acquisition
Improving the quality of the information acquired in the field was a key issue. The former data-acquisition procedure, based on paper forms, generated incomplete and/or inconsistent data. A handheld personal computer equipped with friendly, simple-to-use interface software that automatically guides the user in the filling-in process replaces the paper forms. Also, several consistency checks are performed before accepting information provided by a field technician.
Platforms, applications and data transfers
At the end of each shift, the maintenance/emergency team connects the handheld computer and downloads all of the data to the desktop computer located in the team's regional maintenance center. After executing a local information-processing routine, the data is transmitted to a central database in a remote location (the central database is connected to all the maintenance centers). Software called SisCorrela runs locally on the desktop computers in each maintenance center. Whenever a user opens a new study session, the relevant data is transferred back from the central database, making it available for the local application.
Failure rate evaluation
Given the nature of the data associated with events, incorporating functions for evaluating failure rates into the software was relatively simple. The software can compute failure-rate values considering a single MV circuit or any set of MV circuits defined by the user (including the special case of all MV circuits) in any period of time. The failure rate for cables can be obtained for the whole cable population or segregated by cable type (bare conductor, protected cable, insulated cable or multiplex cable).
The general expression for evaluating the failure rate of cables is:
Where FR
Components for which failure rates can be computed include poles, insulators, transformers, voltage regulators, capacitor banks, arresters, switches, fuse links, automatic reclosers, automatic sectionalizers, energy meters, current transformers, voltage transformers, spacers and connections. Each one of these components can be divided into subtypes according to the type of equipment existing in the distribution network (for instance, the item “arresters” has subtypes “ceramic” and “polymeric”). The general expression for evaluating failure rate of a given component is as follows:
Where FR
The failure rate for components can also be computed taking the total network length as reference (in this case, a formula — see Eq. 1). Figure 1 shows an example of failure rate evaluation per network length; in this case, 700,274 km (435,223 miles). Events that occurred in the whole MV network in 1999 were used in this example.
Number of service operations
The software also enables the number of service operations executed in the network, grouped by component type, to be analyzed.
Reported causes
When opening a new event, the technician reports a cause based on his assessment of the event in the field. The software allows a quantitative analysis of events based on the reported causes, as shown in Fig. 3. It should be noted that for display purposes event causes could be grouped in various ways (environment, maintenance, etc.).
Cost estimates
All events have an initial time and restoration time. The duration of the event multiplied by the corresponding nonsupplied demand — available from the network management system once the operated switches during the event are known — yields the event's nonsupplied energy. The software computes this value and multiplies it by a user-defined unit cost, thus yielding the event's total cost of the nonsupplied energy. The software also computes, based on average unit costs for labor and replaced components, the total cost associated with the two items. Figure 4 shows an example of these calculations.
Estimation of the mean time between failures (MTBF)
For a given period of time, the MTBF is simply computed by dividing the period (in hours or days) by the total number of events. Figure 5 shows the MTBF evaluation for the whole MV network during 1999.
APPLICATION OF THE EVENT INFORMATION SYSTEM
This event information system is now in operation in AES Eletropaulo. The capital investment for this experimental development was minimal, and it includes 15 handheld computers, a desktop computer for each one of the regional operation centers and two-day training programs for technical staff. The benefits of this comprehensive analysis and evaluation of this network management tool are best illustrated by the following two examples:
- Repetitive failures on an MV distribution circuit
Figure 6 shows an analysis tool in which each MV circuit appears along its failure rate (circuits are sorted in descending order of the failure rate).
One circuit with a relatively high failure rate showed an unusually high percentage of events (24%) with reported cause “animals.” One of these events was described in detail by a technician who reported via the handheld computer that in a particular location on the network there was a tee-off position with insufficient phase-to-phase clearance. An on-site visit confirmed the information and also revealed that a passer-by regularly fed pigeons at this location. It was concluded that when the pigeons opened their wings, they caused a phase-to-phase short circuit between two phases, leading to the disconnection of the entire circuit. Therefore, the phase separation was increased, and consideration was taken of the circuit's average maintenance cost, as well as the cost of circuit modifications and an estimated 20% reduction in the circuit's maintenance cost. The payback period for this operation was estimated at 131 days.
- Distribution transformer failure rate
Failure rates for every component within a distribution system must be known prior to any system reliability study, but accurate assessment of these rates is usually a difficult task. While average values are available in the technical literature, they do not take into account the specific features of the distribution system that are under consideration.
This network management system offers a valuable tool for determining component failure rates. Since all events, as well as the reference number of components, are stored in the system on an annual basis, the failure rates and long-term performance is a readily available. The table shows the data relating to distribution transformer failures over a relatively long period, as the events that occurred in 2004 were only available until September 2004; the corresponding failure rate was adjusted to an annual basis.
SUMMARY
The network fault management system now implemented within AES Eletropaulo is a computational system that attempts to establish the cause-effect relationships of events/faults in the operation of a large distribution system. The program offers the opportunity to conduct a number of different analyses that are invaluable for utility staff engaged in the optimization of the financial resources associated with maintenance of the large distribution system. Results confirm the potential available to identify the cause of repetitive faults that would otherwise remain undetected. Also, the information stored in the system's central database will allow for the future development of evaluating other reliability indices, such as System Average Interruption Duration Index (SAIDI) and System Average Interruption Frequency Index (SAIFI).
ACKNOWLEDGEMENT
This article is based on a paper presented by the authors at the 2003 CIRED 17
Dr. Hernán Prieto Schmidt received bachelor's and master's degrees in electrical engineering from Escola Politécnica da Universidade de São Paulo, Brazil, in 1982 and 1989, respectively. In 1994, he received the Ph.D. degree in electrical engineering from the University of London, U.K. Between 1981 and 1990, he worked for E. J. Robba Consultoria & Cia. Ltda., a leading consulting company in São Paulo. Since 1985, Schmidt has been with Escola Politécnica da Universidade de São Paulo. His current research interests include the study of optimization techniques applied to operational and planning problems in distribution systems. hernan@pea.usp.br
Dr. Carlos César Barioni de Oliveira received master's and Ph.D. degrees from Escola Politécnica da Universidade de São Paulo in 1993 and 1997, respectively. Since 1990, he has been with Escola Politécnica da Universidade de São Paulo, where he teaches power system basics and his current research interests include the application of optimization techniques and artificial intelligence in distribution system problems. barioni@pea.usp.br
André Méffe obtained his bachelor's and master's degrees in electrical engineering from the University of São Paulo in 1998 and 2001, respectively. Since 1999, he has worked in the university's department of electrical engineering. His research interests include distribution system planning and the evaluation of technical losses in distribution systems. andre.meffe@daimoninterplan.com.br
Carlos Alexandre de Sousa Penin received bachelor's and master's degrees in electrical engineering from Escola Politécnica da Universidade de São Paulo, Brazil, in 1994 and 2000, respectively. He is currently working toward a Ph.D. degree in electrical engineering. Between 1994 and 1996, Penin worked for Scopus S.A., a leading consulting company in São Paulo, and between 1996 and 1999, he worked for Dresdner Bank Brazil, where he developed computational systems for Internet banking. Between 2000 and 2004, he was with Escola Politécnica da Universidade de São Paulo, where he worked on power system projects concerning investment planning in distribution systems. Since 2005, Penin has worked as a project manager with Daimon Engenharia e Sistemas, a leading consulting company in São Paulo. penin@pea.usp.br
Ivo Teixeira Domingues obtained his bachelor's degree in electrical engineering from UMC-Universidade de Mogi das Cruzes, São Paulo, in 1986. He joined AES Eletropaulo, the electricity utility serving the city of São Paulo. His responsibilities include materials standardization, protective systems and the construction of transmission and distribution lines. ivo.domingues@aes.com
João José dos Santos Oliveira earned his bachelor's degree in electrical engineering from Mackenzie University, São Paulo, and his master's degree in electrical engineering from Escola Politécnica da Universidade de São Paulo in 1997. He has been with AES Eletropaulo since 1978, where he has worked in the areas of planning and operation of overhead and underground distribution. Currently, Oliveira serves as director of engineering. joao.oliveira@aes.com
| Year | Number of Events | Total Number of Transformers | Failure Rate Pu Per Annum |
|---|---|---|---|
| 2000 | 1719 | 156685 | 0.01097 |
| 2001 | 1645 | 159696 | 0.01030 |
| 2002 | 1386 | 162826 | 0.00851 |
| 2003 | 1485 | 162826 | 0.00912 |
| 2004 (Adjusted) | 1235 | 166164 | 0.01035 |
ENERGIZING
Visit www.tdworld.com and sign up to receive the energy-delivery industry's leading e-newsletter, Energizing. Three times a month, you will be informed of the merging trends and news in our industry.
Want to use this article? Click here for options!
© 2008 Penton Media Inc.
